Image Chunker and Verbalization

Image chunking processes image files to index their verbalized content for efficient querying. Currently supported image formats are JPEG (.jpeg, .jpg) and PNG (.png).

Copy link to this section

Image chunking configuration

Image chunking is configured through the Genai.UnstructuredPipeline#parserSettings. The Image Chunker is enabled by default and uses the default-completions client as the default verbalizer.

For detailed configuration options including LLM selection, parsing behavior, and image processing settings, see the Multimodal Parsing guide.

Copy link to this section

Quick configuration example

To change the LLM used for image verbalization:

var pipeline = Genai.UnstructuredPipeline.forId('default-unstructured-pipeline');
pipeline.mergeSettings({
  parserSettings: {
    imageVerbalizationPrompt: { id: 'my_custom_image_prompt' },
  },
});

For more advanced image processing configuration, including resolution settings and verbalization prompts, refer to the Multimodal Parsing documentation.